Verify a decision

{"actor":"judge","decided_at":"2026-08-25T08:26:44.320Z","decision":"review_revise","investigation_id":"a746da55-5f10-4d44-931c-b05b2931a804","new_score":0,"page_slug":"atomic-wallet-hack","prev_score":10,"reason":"The reviewer confirmed 28 of 37 claims, including direct quotations, dollar figures, and North Korea attribution language, with only 1 of 37 (2.7%) actually disputed. Using the spec's disputed+unverifiable formula, the page's disputed_pct is 10.8%, just inside the band that calls for revision rather than approval or denial. The page's litigation section is out of date: claim_findings[31] found the statement that 'no subsequent litigation had been publicly reported in a U.S. court' is stale, since a new class action in the Northern District of California has been publicly reported since mid-2025 with an amended complaint filed in November 2025. That same section also omits a third named individual defendant, Ilia Brusov (claim_findings[29]). Two smaller sourcing errors compound the litigation section's problems: a timeline entry cites the wrong Treasury press release for the Garantex sanction (claim_findings[35]) and another timeline entry gives an incorrect filing date for the original lawsuit (claim_findings[36]). Elsewhere, a loss statistic is mischaracterized as a median when the source reports it as an average (claim_findings[15]), and a wallet-count figure is stated as a single number without disclosing a real spread across sources (claim_findings[1]). None of these touch the page's core allegations -- the hack, the loss scale, or the North Korea attribution -- which held up under review, but the litigation-status error is a substantive currency problem that a reader relying on this page would be misled by, and it aligns with a high-priority coverage gap the reviewer separately flagged. That combination justifies revision even though the aggregate dispute rate is low.","score_delta":-13,"sequence_num":3,"submission_content_hash":null,"submission_id":null,"submission_kind":null,"submission_valence":null,"v":1}